Can someone explain this lineair PSU to me? (and repair)

[obiwanjacobi]

I am working on a repair of a mixer that has this PSU (complete schematic here):


I have never seen a configuration like this before and I am unsure of how it functions. Specifically how the diode bridges are wired up and C's that seem to be in series..? All the U00x parts are either 78xx or 79xx linear regulators (and those I understand).

The problem is that U005 (7924) keeps blowing.  In order to reason about why this could happen I started looking more closely at the circuit and got really confused 

Note: U006 is a 7906 and D006 (zener) is not there, its shorted.
On the left a 2x 2.5V~ (with center tap) on the top and a 2x 14V~ (with center tap => GND) on the bottom come from the transformer.

As you can see in the full schematic the -29V= is used for a vacuum fluorescent display (vu meters) and just pull down a couple of lines to the display. So I figured the problem must be in this power supply...?

Any insights would be most welcome.

[SeanB]

First off replace C007 and C017 with 470uF 63 V capacitors, likely the 25V units are either shot or high ESR. Then replace C004, D005 and D007 with new 1N4007 diodes, and replace the zener and both voltage regulators. Also check R003 and replace C018. Check for any low resistance between the 5VAC rail and ground, and if present disconnect the display board and check again to see if it is the transformer or PSU or the display driver itself.

This uses the capacitors as level shifters, biasing them so they pass the AC power and provide a DC shift to make the different voltage rails from the single 14VAC split supply. Get the capacitors the right way round, they work if incorrect for a short time then go boom. This allows a single power rail that swings symmetrically around 0V to generate a 035V rail for the display, and at the same time make a 28V rail to drive the phantom power side, using the bridge rectifier D002 for the phantom rail and D004/5 for the negative rail.

While you are there replace C008 and C009 with the same 63V capacitors before you kill the 24V phantom power rail, likely the caps used are marginal.

[tatus1969]

The designer has probably done this to be able to live with a transformer with only two secondary windings. The main bridge rectifier is D001 and it is configured to produce a bipolar output that is referenced to GND.

The output voltage of U004 seems to be quite high (he is using 50V electrolytics there), which would require both secondary windings in series. But D001 is preventing this already. So he configured D002 to work as a dual charge pump in combination with C008 / C009.

This can be easier to understand when you look at C007/D004/D005. They are working in the same way. 1) positive transformer voltage charges the capacitor via D004 2) negative phase discharges it into C017 via D005.

To U005: I can imagine three reasons why it keeps blowing:
1) input overvoltage or reversal
2) output overvoltage or reversal
2) current flowing in the wrong direction at powerup or -down

Overtemperature is unlikely because they are internally protected.

Maybe check or replace C017 / C018, that would be my first guess.

[obiwanjacobi]

Thank you both.

This uses the capacitors as level shifters, biasing them so they pass the AC power and provide a DC shift to make the different voltage rails from the single 14VAC split supply. Get the capacitors the right way round, they work if incorrect for a short time then go boom. This allows a single power rail that swings symmetrically around 0V to generate a 035V rail for the display, and at the same time make a 28V rail to drive the phantom power side, using the bridge rectifier D002 for the phantom rail and D004/5 for the negative rail.

How does low ESR affect this? I would have guessed that a low ESR may degrade the efficiency and the voltage levels, but I still cannot explain how it can blow the 7924..?

Then replace C004, D005 and D007 with new 1N4007 diodes, and replace the zener and both voltage regulators.

You mean the zener D006 and the U005/U006 regulators? The zener is not there, it is a wire link.

To U005: I can imagine three reasons why it keeps blowing:
1) input overvoltage or reversal
2) output overvoltage or reversal
3) current flowing in the wrong direction at powerup or -down

1) can that happen with bad caps? (high ESR)
2) I still don't get how that output works, it is tied to the center tap of the 2x 2.5V transformer output...?
3) Could it be some kind of discharge from the fluorescent display. The 100k pull-down R's are all good - measured it.

[SeanB]

Zener is needed because nobody makes a high voltage capable negative voltage regulator, and the supply probably is at around -35V, which is going to be outside the design ratings of many voltage regulators, so the zener diode is there to reduce it slightly so the regulator does not blow. Works with the original regulator but other manufacturers parts ( or even different batches) might not be so tolerant of high input voltage, so put a 1W zener in there. ESR of the caps is not an issue so long as it is low, the transformer DC resistance and AC reactance dominates, but failing capacitors with high ESR will lead to high ripple on the rails, and this can fail the regulator from the input dropping below the output voltage every cycle.

The output is there to bias the display cathode at -24V or so, and the 5VAC is to drive the directly heated cathodes of the display.

[obiwanjacobi]

Thanx for that clear explanation, makes sense. The datasheet of the 7924 part I used said -40V max input, but I don't see why I should drive it so hard if it is not necessary and this leaves some wiggle room.

I've ordered the parts and will continue this repair when they get in.
U006 (7906) is currently a small transistor package (don't know those names by heart) but I cannot get those, so I order the TO-220 1A version. I think (hope) it will fit...

I have to be very careful, the PCB doesn't take much rework before tracks/pads are lifting...

[obiwanjacobi]

I replaced all the suggested components, all but the U006 (79L06) thinking it did work and might not need replacing. U005 (7924) still got hot even with the now fitted zener diode (5.6V). So I replaced U006 and U005 now gets a bit warm but nothing to be concerned about. I test this on the bench by connecting the display (nothing else) and leaving it on for 30 minutes.

I ran the caps I replaced through the tester and they seemed good - although one may have been dodgy, it took 3 tries to get a good reading. So perhaps it was just that pesky U006 all along... 

I am now working on the MIDI controller (know as 8200?) that is also fitted in this unit and controls the channel mutes. There does not seem to be any documentation on it...   

Thanks for the guidance.

[tatus1969]

good to hear

[obiwanjacobi]

Here's an update for those interested.

The PSU seems to function correctly now for a while, however the +12V and -12V regulator get very warm as well as the diode bridge that is before that.

The 7812 and 7912 where plastic housings that do not require any isolation to the heat sink. I ordered new regulators with metal plats (TO-220) and put those in (isolated only the 7912). The temp difference between the regulator package and the heat sink is very low, way lower than it was before but it still gets pretty warm.

(same image as in first post)

I have the top and bottom connectors disconnected and only connected the middle connector which feeds the channel strips. With only half the channels fitted (10) my multi meter said only about 100mA was drawn on the +12V and the -12V lines. I retested with a 1.5 ohm resistor as a shunt (sorry, lowest I could find in my parts bin) and came out a little higher (250mV over 1.5Ohm =166mA) but well under 200mA on both +/-12V power rail lines. Both with my multimeter as with the resistor in series the heat generated was significantly less.

For 10 channels a current of say 160mA does not sound way off. That would mean we stay below 0.5A for the entire device. I cannot imagine that the 160mA I am drawing would heat up these component so much.

I have looked at the rectified levels with a scope and it has a ripple of about 0.5V (p2p) at 20V (instead of the 16V specified in the schematic) - with a typical saw-tooth like wave form of the slowly discharging caps. The regulated +12V and -12V are very flat and I could not measure any ripple.

I did not replace any of the caps in this part of the circuit and am thinking of doing that. I am not sure how this much heat is generated on a relatively small current draw...


As for the MIDI controller, its processor gets very hot when powered up and it does nothing. A new processor is $30,=. So I have given up on that module and removed it from the unit and restored all the channel strips to let the mute work stand alone again.